To train or evaluate MedViT models on 17 medical datasets, follow this "Evaluation".
You can find a tutorial for visualizing the Grad-CAM heatmap of MedViT in this repository "visualize".
Below is the performance summary of MedViT on various medical imaging datasets.
🔹 Model weights will be available soon.
| Dataset | Task | Overall Accuracy (%) |
|---|---|---|
| PAD-UFES-20 | Multi-Class (6) | 63.6 |
| ISIC2018 | Multi-Class (7) | 77.1 |
| Fetal-Planes-DB | Multi-Class (6) | 95.3 |
| CPN X-ray | Multi-Class (3) | 98.2 |
| Kvasir | Multi-Class (8) | 82.8 |
| ChestMNIST | Multi-Class (14) | 96.3 |
| PathMNIST | Multi-Class (9) | 95.9 |
| DermaMNIST | Multi-Class (7) | 78.1 |
| OCTMNIST | Multi-Class (4) | 92.7 |
| PneumoniaMNIST | Multi-Class (2) | 95.1 |
| RetinaMNIST | Multi-Class (5) | 54.7 |
| BreastMNIST | Multi-Class (2) | 88.2 |
| BloodMNIST | Multi-Class (8) | 97.9 |
| TissueMNIST | Multi-Class (8) | 69.9 |
| OrganAMNIST | Multi-Class (11) | 95.8 |
| OrganCMNIST | Multi-Class (11) | 93.5 |
| OrganSMNIST | Multi-Class (11) | 82.4 |
MedViT is released under the MIT License.
💖🌸 If you find my GitHub repository useful, please consider giving it a star!🌟
@article{manzari2025medical,
title={Medical Image Classification with KAN-Integrated Transformers and Dilated Neighborhood Attention},
author={Manzari, Omid Nejati and Asgariandehkordi, Hojat and Koleilat, Taha and Xiao, Yiming and Rivaz, Hassan},
journal={arXiv preprint arXiv:2502.13693},
year={2025}
}
@article{manzari2023medvit,
title={MedViT: a robust vision transformer for generalized medical image classification},
author={Manzari, Omid Nejati and Ahmadabadi, Hamid and Kashiani, Hossein and Shokouhi, Shahriar B and Ayatollahi, Ahmad},
journal={Computers in Biology and Medicine},
volume={157},
pages={106791},
year={2023},
publisher={Elsevier}
}